revert example script

Signed-off-by: George Ohashi <[email protected]>

examples/quantization_w4a16/llama3_example.py

@@ -1,105 +1,80 @@
-# from datasets import load_dataset
-# from transformers import AutoModelForCausalLM, AutoTokenizer
-
-# from llmcompressor.modifiers.quantization import GPTQModifier
-# from llmcompressor.transformers import oneshot
-
-# # Select model and load it.
-# MODEL_ID = "meta-llama/Meta-Llama-3-8B-Instruct"
-
-# model = AutoModelForCausalLM.from_pretrained(
-#     MODEL_ID,
-#     device_map="auto",
-#     torch_dtype="auto",
-# )
-# tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
-
-# # Select calibration dataset.
-# DATASET_ID = "HuggingFaceH4/ultrachat_200k"
-# DATASET_SPLIT = "train_sft"
-
-# # Select number of samples. 512 samples is a good place to start.
-# # Increasing the number of samples can improve accuracy.
-# NUM_CALIBRATION_SAMPLES = 512
-# MAX_SEQUENCE_LENGTH = 2048
-
-# # Load dataset and preprocess.
-# ds = load_dataset(DATASET_ID, split=DATASET_SPLIT)
-# ds = ds.shuffle(seed=42).select(range(NUM_CALIBRATION_SAMPLES))
-
-
-# def preprocess(example):
-#     return {
-#         "text": tokenizer.apply_chat_template(
-#             example["messages"],
-#             tokenize=False,
-#         )
-#     }
+from datasets import load_dataset
+from transformers import AutoModelForCausalLM, AutoTokenizer
 
+from llmcompressor.modifiers.quantization import GPTQModifier
+from llmcompressor.transformers import oneshot
 
-# ds = ds.map(preprocess)
+# Select model and load it.
+MODEL_ID = "meta-llama/Meta-Llama-3-8B-Instruct"
 
+model = AutoModelForCausalLM.from_pretrained(
+    MODEL_ID,
+    device_map="auto",
+    torch_dtype="auto",
+)
+tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
 
-# # Tokenize inputs.
-# def tokenize(sample):
-#     return tokenizer(
-#         sample["text"],
-#         padding=False,
-#         max_length=MAX_SEQUENCE_LENGTH,
-#         truncation=True,
-#         add_special_tokens=False,
-#     )
+# Select calibration dataset.
+DATASET_ID = "HuggingFaceH4/ultrachat_200k"
+DATASET_SPLIT = "train_sft"
 
+# Select number of samples. 512 samples is a good place to start.
+# Increasing the number of samples can improve accuracy.
+NUM_CALIBRATION_SAMPLES = 512
+MAX_SEQUENCE_LENGTH = 2048
 
-# ds = ds.map(tokenize, remove_columns=ds.column_names)
+# Load dataset and preprocess.
+ds = load_dataset(DATASET_ID, split=DATASET_SPLIT)
+ds = ds.shuffle(seed=42).select(range(NUM_CALIBRATION_SAMPLES))
 
-# # Configure the quantization algorithm to run.
-# #   * quantize the weights to 4 bit with GPTQ with a group size 128
-# recipe = GPTQModifier(targets="Linear", scheme="W4A16", ignore=["lm_head"])
 
-# # Apply algorithms.
-# oneshot(
-#     model=model,
-#     dataset=ds,
-#     recipe=recipe,
-#     max_seq_length=MAX_SEQUENCE_LENGTH,
-#     num_calibration_samples=NUM_CALIBRATION_SAMPLES,
-# )
+def preprocess(example):
+    return {
+        "text": tokenizer.apply_chat_template(
+            example["messages"],
+            tokenize=False,
+        )
+    }
 
-# # Confirm generations of the quantized model look sane.
-# print("\n\n")
-# print("========== SAMPLE GENERATION ==============")
-# input_ids = tokenizer("Hello my name is", return_tensors="pt").input_ids.to("cuda")
-# output = model.generate(input_ids, max_new_tokens=100)
-# print(tokenizer.decode(output[0]))
-# print("==========================================\n\n")
 
-# # Save to disk compressed.
-# SAVE_DIR = MODEL_ID.split("/")[1] + "-W4A16-G128"
-# model.save_pretrained(SAVE_DIR, save_compressed=True)
-# tokenizer.save_pretrained(SAVE_DIR)
+ds = ds.map(preprocess)
 
 
-from transformers import AutoModelForCausalLM, AutoTokenizer
+# Tokenize inputs.
+def tokenize(sample):
+    return tokenizer(
+        sample["text"],
+        padding=False,
+        max_length=MAX_SEQUENCE_LENGTH,
+        truncation=True,
+        add_special_tokens=False,
+    )
 
-from llmcompressor import oneshot
-from llmcompressor.modifiers.quantization import QuantizationModifier
 
-# Define the model to compress
-MODEL_ID = "meta-llama/Meta-Llama-3-8B-Instruct"
+ds = ds.map(tokenize, remove_columns=ds.column_names)
 
-# Load the model
-model = AutoModelForCausalLM.from_pretrained(
-    MODEL_ID, device_map="auto", torch_dtype="auto"
-)
-# Load the tokenizer
-tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
+# Configure the quantization algorithm to run.
+#   * quantize the weights to 4 bit with GPTQ with a group size 128
+recipe = GPTQModifier(targets="Linear", scheme="W4A16", ignore=["lm_head"])
 
-# Define the recipe, scheme="FP8_DYNAMIC" compresses to W8A8, which is
-# FP8 channel-wise for weight, and FP8 dynamic per token activation
-recipe = QuantizationModifier(
-    targets="Linear", scheme="FP8_DYNAMIC", ignore=["lm_head"]
+# Apply algorithms.
+oneshot(
+    model=model,
+    dataset=ds,
+    recipe=recipe,
+    max_seq_length=MAX_SEQUENCE_LENGTH,
+    num_calibration_samples=NUM_CALIBRATION_SAMPLES,
 )
 
-# compress the model
-oneshot(model=model, recipe=recipe)
+# Confirm generations of the quantized model look sane.
+print("\n\n")
+print("========== SAMPLE GENERATION ==============")
+input_ids = tokenizer("Hello my name is", return_tensors="pt").input_ids.to("cuda")
+output = model.generate(input_ids, max_new_tokens=100)
+print(tokenizer.decode(output[0]))
+print("==========================================\n\n")
+
+# Save to disk compressed.
+SAVE_DIR = MODEL_ID.split("/")[1] + "-W4A16-G128"
+model.save_pretrained(SAVE_DIR, save_compressed=True)
+tokenizer.save_pretrained(SAVE_DIR)